Democracy Counts

The Transparency Project in 18 Minutes

2011-12-20T11:58:00.000-08:00

Donald Regalmuto has put together an edit of my presentation in Marin County in October 2010. (It even includes my first ever attempt at animation, done with the open source program Synfig Studio. Even though it's extremely basic and lasts only about 15 seconds, I'm really happy to have animated something.) The video is about 18 minutes and may be of interest and/or use. This was put together thanks to Lori Grace.

If the version below is clipped, you can see it directly at You Tube:

http://www.youtube.com/watch?v=z8Zxw2Im5io&feature=youtu.be

Updated November 2011 Election Results

2011-11-20T11:23:00.001-08:00

After I manually scanned through all ballot images to search for and make any corrections to the TEVS results, I located 35 vote targets whose software interpretations needed correction. Mostly, these were due to extremely light check marks. The scan of the 12,000 images, with votes overlaid on them by the software, went at about 90 to 100 images per minute, or 90 to 100 ballots per minute, since these were single sided ballots.

Having made those corrections, my independent results off of the Transparency Project's images match the county's Final Unofficial Results exactly except for the following:

Dave Saunderson 1559 versus county report of 1557
Emil Fierabend 441 versus county report of 442
John W Corbett 1377 versus county report of 1376
Joe O'Hara 104 versus county report of 102
Judy Gower 162 versus county report of 164
Mike Seeber 55 versus county report of 54
Susan Johnson 3104 versus county report of 3103
Zachary Thoma 16 versus county report of 17

As my interpretation of votes may not have followed the official rules, it is very reasonable to assume that the errors are in the independent count. None of the discrepancies would change any results.

The only reason I am not calling this "entry" final is that we may have an opportunity to do multiple-person recounting of the races with discrepancies.

Preliminary Results, Nov 2011 Elections

2011-11-17T10:57:00.000-08:00

Given the dismal 22% turnout, there were only 12,269 ballots for the Transparency Project to scan. We began with 3.5 hours of scanning on Veteran's Day, and then 3 shifts this week totaling 11 more hours.

I ran the scans through TEVS yesterday and have manually counted the (fewer than 10) ballot scans it rejected. I also checked for votes on about 40 ballot images when a vote region was indicated as just a smidgen above the darkness cutoff. The results are still missing for one ballot, where we mistakenly scanned the blank reverse. That ballot will have to be retrieved from the County's ballot storage.

The results compare well with the Final Unofficial Results posted at the County web site.

I'll point to a spreadsheet once I've been able to check the results more thoroughly, but here
are screen shots of the spreadsheet in its current form. The write-in results are incomplete, due to my mistake, but the other results are mostly within 1 of the Final Unofficial Results, with a couple off by 2.

Note that some candidates have more than one line, where optical character recognition generated multiple versions of their name and I did not merge them (many such variants have already been merged in software -- the ones that show are the ones I missed).

If I can shop online, why can't I vote online?

2011-11-01T12:13:00.001-07:00

An important explanation of a common question, by David Jefferson, at the Election Law Blog.

http://electionlawblog.org/?p=24849

Progress towards a graphical UI for TEVS

2011-10-20T08:04:00.000-07:00

I'd hoped to have a tested graphical user interface version of TEVS ready before November, but that has proven impossible. As I won't be able to touch this for some time, I've posted the existing code in an alternate repository, tevs.gui, at tevs.googlecode.com. Though it works soup-to-nuts for scanning, processing, counting, and reporting at my setup, it is untested, undocumented, and incomplete.

Here are screenshots, from which behavior can be inferred. Clicking the images will get you larger versions.

There are three control panel tabs, corresponding to the stages of the process.

First tab on control panel is for scanning. The GUI calls out to a separate scanning process, which gets the scans and writes them to files. The GUI then reads the files. Controls allow for choosing endorser if available, changing resolution, selecting duplex/simplex and ballot size. If a scanner is not found (unplugged, bad connection) the user can check and click "Search for scanner" to connect. If it is present, its type is displayed. More than one scan source is not now handled.

Second tab is for processing scanned ballots into a vote database, getting info from the database, signing and writing all info to a DVD. The signing and writing process is done by prompting a user through a terminal window dialog with tar, gpg, etc..., and encouraging them NOT to do the signing on the actual machine running TEVS.

Decisions about how to insist on an educated user have been the most difficult part of this process -- it is pointless to produce a tool that is simply trusted. The disk's user is free to alter the database at will -- the assumption here is that the group using the tool is independent of the elections office and trusts itself

Merging OCR variants -- templates are built automatically on supported vendor ballot designs when the system encounters new ballot types (as in new precincts, for example), but the OCR in building the templates will sometimes read the same contest or choice with different errors from one precinct to the next. The system makes its best guesses and then asks you to confirm or alter associations between different variants of contests and choices before it totals things up. To ignore this, all one needs to do is click "Done" and the system skips on to overvote processing and counting up the votes.

Showing results -- they can also be put into a PDF and printed.

Third tab is for displaying results on ballots. You can walk through ballots sequentially. You can also query the database for ballots with particular characteristics (ambiguous votes, overvotes, particular precinct, etc...), and can click through the resulting list to see those ballots.

The user interface is done using Gtk, and its appearance is generated via Glade, a user interface design tool. This makes it easily alterable and translatable. Tasks like scanning, database access, and ballot processing are farmed out to other processes, generally slight modifications of the non-UI routines that already exist.

Scotia August 2011 Results

2011-09-02T10:22:00.000-07:00

We've scanned the 151 ballots for Scotia's August 2011 election and counted them with TEVS. The results are below.

Interestingly, because one voter voted for three candidates and then wrote them in, marking three write-in boxes, that voter's votes in the Director contest were considered invalid -- only five boxes can be selected, and if more than five are selected, none are counted. That ballot, with the voter's handwriting removed, is the image you see above.

Because this is such a small set of ballots, we've chosen to hide the handwriting of the write-in votes by modifying the few scans with write-ins to erase the handwriting and replace it with standard text. The scans are here (29 megabytes), the digital signature to validate the scans is here, a screenshot of TEVS showing the first ballot is here, and the detailed spreadsheets are here (as Open Office) and here (as Excel).

(Note: "ambiguous" in this context doesn't mean the voter was ambiguous, but that TEVS had contradictory results on the two different tests it uses to decide if a mark is enough to register as a vote. Such votes are hand-checked.)

count | choice_text
-------+-----------------------
68 | CAROLYN DEPUCCI
70-1 | RICK WALSH (one vote labeled ambiguous inspected and removed from ballot 77)
96 | JAMES BARNES
48 | MARILYN SANDERSON
109 | JOHN BROADSTOCK
140 | YES
69 | KEVIN LALOLI
40 | WILLIAM BILL STEPHENS
103 | GAYLE MCKNIGHT
39 | JOHN CANESSA
9 | NO
7 | Writein
(12 rows)

Not included above due to "overvote" in contest:

substring | choice_text
--------------------+-----------------------
roc/001/001067.jpg | GAYLE MCKNIGHT
roc/001/001067.jpg | JAMES BARNES
roc/001/001067.jpg | JOHN CANESSA
roc/001/001067.jpg | MARILYN SANDERSON
roc/001/001067.jpg | WILLIAM BILL STEPHENS
roc/001/001067.jpg | RICK WALSH
roc/001/001087.jpg | GAYLE MCKNIGHT
roc/001/001087.jpg | JAMES BARNES
roc/001/001087.jpg | JOHN CANESSA
roc/001/001087.jpg | MARILYN SANDERSON
roc/001/001087.jpg | WILLIAM BILL STEPHENS
roc/001/001087.jpg | RICK WALSH
roc/001/001088.jpg | CAROLYN DEPUCCI
roc/001/001088.jpg | JOHN BROADSTOCK
roc/001/001088.jpg | MARILYN SANDERSON
roc/001/001088.jpg | Writein
roc/001/001088.jpg | Writein
roc/001/001088.jpg | Writein
roc/001/001026.jpg | JAMES BARNES
roc/001/001026.jpg | Writein
roc/001/001026.jpg | Writein
roc/001/001026.jpg | Writein
roc/001/001026.jpg | Writein
roc/001/001026.jpg | Writein
(24 rows)

A Voting System without "Spoilers" -- Approval Voting

2011-07-15T09:48:00.000-07:00

A perpetual problem with our current voting system is that it subjects third-party candidates to charges that they are "spoilers" while forcing voters to vote strategically rather than honestly. That is, if you really like candidate C, but realize that the most likely winner is either candidate A or candidate B, you may feel obligated to vote for A to prevent a win by B.

How hard is it to get around this problem? Not hard at all -- it just means considering the system that was used to vote for the first four U.S. presidents. That's Approval Voting.

Under approval voting, you can vote for as few or as many choices as you like in each election. Then, every vote for every candidate is added up, and the candidate with the most votes (approvals) wins. Easy.

More here, at the Center for Election Science.

TEVS Live CD now online

2011-06-07T12:36:00.000-07:00

One of the most frustrating things about the TEVS software has been its dependence on a modified version of an open source image utility library. It was an early mistake but, because it worked for me on my setup, it wasn't until very recently that I removed the dependency on this modified version. The dependency made it needlessly difficult for others to build TEVS. But it's gone now.

In addition, a lot of cleanup, documenting, and restructuring of the ballot handling code has been done by James Frasche, thanks to funding provided by Lori Grace and the Grace Institute for Democracy and Election Integrity. (Thanks!) This will make it easier for people to add handling for new ballot types going forward.

So it finally makes sense to make a Live CD available to those who might want to have an environment in which they can run the code, figure out what they want to add or improve, and just add and improve it.

The Live CD is a version of the Ubuntu Live CD which includes TEVS, all the programs it uses, a few sample ballot images, and a link to the code repository at tevs.googlecode.com. While it's NOT slick, it does provide a few buttons that allow you to run TEVS on a small set of provided ballot images and see the results, see the XML templates that get built, then run it again on the same images to re-use the templates, then run it on the images at a lower resolution, to see the difference in speed between 300 dpi images and 150 dpi images. Because the Live CD's 900 meg size won't fit on a CD, it needs to be burned onto a DVD. Once you've burned it to a DVD, though, you can just pop it into any PC or modern Mac and run Ubuntu and TEVS without installing them on your machine.

If you've never used Ubuntu Linux, don't worry. It should be perfectly straightforward how to use things for the demo. You can also see how easy it is to use Ubuntu for things like web browsing and general office work.

The Live CD should be a good tool for demonstrating the possibilities TEVS provides. It can be downloaded from this link: http://dl.dropbox.com/u/18212385/customdist.iso

TEVS Source Code Update Online

2011-03-21T17:46:00.000-07:00

A Mercurial repository of the TEVS source code is now online at tevs.googlecode.com.

The entry point for the data extraction code is at tevs/main.py, and the generic ballot handling code is at tevs/Ballot.py. The display code, in tevs/TEVS.py, will be undergoing major modifications.

The reason for putting the code online at this point is not that it's ready for prime time, but that the data extraction part has now been restructured to the point that it's no longer mostly a waste of time to use it as a starting point.

Happy Holidays

2010-12-24T09:19:00.001-08:00

May the human spirit prevail.

(n.b.: Jonathan is no relation, nor is the composer.)

Some Preliminary Results for November 2010

2010-12-01T15:09:00.001-08:00

I've run the set of Transparency Project scans through the counting software and have preliminary results. More precise results will need to wait until I've had an opportunity to go through things much more carefully, but I feel comfortable releasing the percentages I've calculated for some of the races.

For some close races which had substantial interest, here are my preliminary percentages compared with the official results. Please keep in mind that the variances of 1/10 of a percent are very reasonable given that my count still needs to properly resolve over-votes -- where there were too many votes in a contest -- and is still missing approximately 150 sheets which were refused by the counting software. There are also a small number of very-lightly-marked ballots from which a very few additional votes will be harvested, but these should have little effect on the percentages.

DA
Gallegos 52.0 vs county 52.08
Jackson 47.7 vs county 47.61

Eureka Ward 3
Newman 44.3 vs county 44.18
Kuhnel 42.0 vs county 41.95
Manns 13.5 vs county 13.59

Supervisor District 5
Cleary 49.1 vs county 49.04
Sundberg 50.5 vs county 50.38

Progress Update 4

2010-12-01T08:02:00.000-08:00

Complete sets of ballot scans are now available from the Humboldt County Elections office. Even when compressed, the data takes up six DVDs, so it is more than I plan on posting to the 'net (others might, though). The uncompressed, higher quality images take up something like a dozen DVDs.

As I mentioned in an earlier update, our percentages track the official percentages, which is what you'd expect. I will post my calculations of initial race percentages here, probably later today, so any interested parties can compare them with the official results.

As the Transparency Project only received the very last of the ballots on Tuesday, I don't plan on posting absolute numbers until at least a few days from now, after I've had a chance to rerun counting software, go through ambiguous votes, and look at some of the 350 images (out of 207,000+) that didn't get processed automatically.

Progress Update 3

2010-11-28T18:45:00.001-08:00

Humboldt County released the remaining ballots for us to scan late last week. As of Sunday night, we have fewer than 1500 ballots remaining to be scanned (each ballot is two sheets). That means we're 97% done with the scanning.

We hope to have our preliminary counts available by Tuesday, when the elections office certifies their results. Our incomplete results do not disagree with the official results in who has won and who has not.

Progress Update 2

2010-11-17T08:00:00.000-08:00

Scanners will complete the initial set of ballots later today. We will now have to wait until the Elections Department is ready to provide us with the remaining ballots, roughly 12,000. That will likely take a week or so, after which it should take only two days of further scanning.

Progress Update

2010-11-12T17:09:00.000-08:00

Scanners have now reached the 100,000 images count for the November 2010 election.

Because this election used a ballot with two sheets, that represents 25,000 complete ballots.

Frightening Experiment with Touch Screen Voting

2010-10-29T08:25:00.000-07:00

Professor Doug Jones of the University of Iowa experimented with setting up a touch screen voting machine to flip votes for Obama and McCain. He then had people "cast their votes" on the machine, and interviewed them about the experience.

Click here for complete paper.

One thing we did in our experiment was to ask a number of user satisfaction questions as a followup. An important question in this batch was: "Are you confident that the machine correctly recorded your vote?" The majority of voters were very confident. Curiously, this confidence did not go down much when we rigged the machine to flip the votes for McCain and Obama.

Why? First, note that half the voters did not notice this switch (emphasis added). People felt strongly about their presidential preference, but people, in general, are not very good at proofreading. The summary screen that comes up at the end of the voting session clearly showed McCain for the Obama voters and visa versa, yet they did not notice.

Among people who did notice, the usual reaction was "Huh?" and then they went back to fix the problem. In general, people assume that the machine is right and assume that it was their error, not the machine's dishonesty. Only a small fraction of our voters
(or more properly, experimental subjects) commented on the fact that we'd flipped their votes.

This has important consequences for the real world. Most people do not complain when there is a problem. They assume that it was their mistake and go back and fix it. I assume that things would be different if the machine did not let them fix the vote flip, but if the machine lets them fix it, the fact that some voters are complaining suggests that there may be far more voters who notice the problem and are silent, and that even more voters may have had their votes flipped but didn't notice the problem.

Absolutely Devastating

2010-10-08T16:25:00.000-07:00

Absolutely Devastating Testimony on Internet Voting

Prof. J. Alex Halderman of the University of Michigan describes his group's ability to take complete control of Washington, DC's test of internet voting security.

Internet Voting comments by David Jefferson of Verified Voting

2010-10-06T12:26:00.001-07:00

(Feel free to forward in full with attribution to David Jefferson of Verified Voting.)

University of Michigan Prof. Alex Halderman has now released some details about his successful attack on the District of Columbia's proposed Internet voting system which has been under test for the last week. (See www.freedom-to-tinker.com.) It is now clear that Halderman and his team were able to completely subvert the entire DC Internet voting system remotely, gaining complete control over it and substituting fake votes of their choice for the votes that were actually cast by the test voters. What is worse, they did so without the officials even noticing for several days.

Let there be no mistake about it: this is a major achievement, and supports in every detail the warnings that security community have been giving about Internet voting for over a decade now. After this there can be no doubt that the burden of proof in the argument over the security of Internet voting systems has definitely shifted to those who claim that the systems can be made secure.

Computer security and election experts have been saying for over 10 years that the transmission of voted ballots over the Internet cannot be made safe with any currently envisioned technology. We have been arguing mostly in vain that:

1) Remote attack: Internet voting systems can be attacked remotely by any government, any criminal syndicate, or any self aggrandizing individual in the world.

2) Effective defense virtually impossible: There are innumerable modes of attack, from very easy to very sophisticated, and if anyone seriously tried to attack an Internet election the election officials would have essentially no chance at successfully defending. The election would be compromised

3) Attackers may change votes arbitrarily: An attack need not just prevent people from voting (bad as that would be), but could actually change large numbers of votes, allowing the attackers to determine the winner.

4) Attacks may be undetected: An attack might go completely undetected. The wrong people could be elected and no one would ever know.

Prof. Halderman demonstrated all of these points:

1) Remote attack: His team of four conducted their attack remotely, from Michigan, via the Internet, without ever getting near Washington, D.C.

2) Effective defense virtually impossible: Although they were restricted from most modes of attack (which would be illegal even in this test situation), they still succeeded in completely owning (controlling) the voting system within about 36 hours after it was brought up, even though they had only 3 days of notice of when it would start. They happened to use one particular small vulnerability that they identified, but they are quite confident that they could have penetrated in other ways as well. Most likely they were the only team to even attempt to attack the system seriously; yet in a real election with something important at stake multiple teams might attack. The fact that the only team that even tried succeeded so quickly is a demonstration lots of other groups from around the world could also have done it.

3) Attackers may change votes arbitrarily:They not only changed some of the votes, they changed them all, both those cast before they took control of the system and those cast afterward. There is no way that officials can restore the original votes without the attackers' help.

4) Attacks may be undetected:The attack was not detected by the officials for several days, despite the fact that they were looking for such attacks (having invited all comers to try) and despite the fact that the attackers left a "signature" by playing the Michigan Fight song after every vote was cast!

This successful demonstration of the danger of Internet voting is the real deal. It doesn't get any better than this, people.

Alex Halderman, his graduate students Eric Wustrow and Scott Wolchok, and their colleague Dawn Isabel, all deserve enormous credit, congratulations, and thanks.

Scanning Completed, Results Match Well

2010-07-03T16:05:00.000-07:00

The county finished processing its last ballots last Thursday, and the transparency project scan was completed yesterday. Preliminary results for a few races show county vote percentages and independent vote percentages varying by no more than 0.05% (one vote in two thousand). The independent results are still missing about fifteen ballots that could not be processed automatically, and there may be other minor adjustments.

No contests are at issue.

It may be a while before the scans are available. For many precincts, only one or two ballots were cast for some parties. Because these ballots could be used to identify an individual voter, we will need to remove them before releasing the set of scans.

Scanning Update

2010-06-23T10:17:00.000-07:00

The project's volunteer scanners caught up last week with available ballots.

We will scan the last ballots as they are made available to us from the elections office, which is still doing operations like checking provisional ballots.

Ballot scans and an independent count should be available around the end of June.

June 2010 Primary

2010-06-13T12:05:00.001-07:00

The transparenteers started scanning last Tuesday's ballots on Thursday. Some folks came in on Saturday, as well. Twelve thousand are now scanned, and another sixteen thousand should be scanned by mid-week. We will then pause until the Elections Department makes available the final batch of seven or eight thousand. I believe these are composed mostly of provisionals and absentee ballots received close to the election.

These are 8.5 x 17 ballots. Although the scanner processes them at a rate of 2,500 or so double sided ballots per hour, we're finding a typical throughput of 1,000 per hour.

Two Legal 'Situations'

2010-02-20T09:23:00.000-08:00

These news items aren't directly related to the Humboldt County Election Transparency Project, but may be of interest to those who are interested in election integrity.

First, New York state recently awarded a voting machine contract to the nation's largest voting machine supplier, ES&S. According to Dominion, another bidder, New York originally ranked Dominion's solution higher. Dominion was also a substantially lower priced bidder. But New York then raised ES&S' ranking after the company added an "ease of use" feature that would, according to Dominion, be illegal in New York due to its compromise of security. Information can be found at Bo Lipari's blog -- http://www.bolipari.com/boblog/2010/02/dominion-sues-to-stop-new-york-city-contract-with-ess/.

Also this morning, news from Clay County, Kentucky, where, according to the Lexington Herald Leader, a former precinct worker has testified in court about how she stole votes from voters using voting machines. She testified that she was instructed in her technique by County Clerk Freddy Thompson, the chief election officer in that county.

The vote-buyers took advantage of some confusion caused by new voting machines the county had that year, White said. The machines had a "Vote" button that people could push to review their choices, then a second button they had to push to record the choices and finish voting.

Meanwhile, here in Humboldt County, CA, we use optically scanned paper ballots with no helpful "Vote" buttons to push but not vote. And, with the cooperation and assistance of County Clerk Carolyn Crnich, the Transparency Project has made available independent scans of every ballot cast in the last three elections, enabling independent recounts.

Cybergate article by Simon Worrall

2010-01-18T10:36:00.000-08:00

I'll be very interested in the story this well-respected reporter has in the newest Maxim magazine. I haven't read it yet, but here's his blog entry:

http://simonworrallauthor.com/?p=186

Completely Unofficial Count

2009-12-03T19:04:00.000-08:00

My independent count of the November 3rd election is now up at http://www.humetp.org, along with the ballot montage files. The independent count is not identical to the official count, but it is close enough to support the accuracy of the official count. (It shouldn't be too surprising when different machines come up with numbers that vary by one or two -- they may have different thresholds for deciding when a mark is a vote.)

I've also posted a draft of my report on the independent counting process: it's at http://docs.google.com/View?id=dgxf37nt_155rw4spndw

Count Status

2009-12-02T08:06:00.000-08:00

Because the project was mentioned in today's Eureka Times-Standard, I want to note that we've essentially completed an independent count of the November 3rd election. It matches the results from the County quite well -- no surprises whatsoever -- but we want to dot our i's and cross our t's before making it public.

Scans Available

2009-11-29T15:42:00.000-08:00

The scanning from the November 3rd election was completed about a week ago. Montage files of the scans are now available at http://www.humetp.org/20091103 .

We should be releasing an independent count of the election within a few days.

November 3rd election, scanning in progress

2009-11-12T16:35:00.000-08:00

We've started scanning ballots from the November 3rd election. Due to the horribly light turnout, we're almost done.

This election is using Hart ballots and equipment, rather than Diebold/Premier.

A couple of technical items...

First, the Hart ballots are rounded on the upper left corner (when viewing the front side). This makes it much easier to ensure that all ballots are properly oriented. Second, the Hart ballots are on a paper stock that seems about the same weight as office paper, while the Diebold/Premier were on card stock. The regular paper weight has substantially reduced our mis-feed rate.

I'm hoping to be able to run Ballot Browser to generate independent counts early next week.

May 19 Special Election: official and independent counts

2009-06-15T18:32:00.000-07:00

The May 19 Special Election final results are in.

The County results are from the Humboldt County Elections office website.

The independent results are generated by running Ballot Browser against the Transparency Project's ballot images, using a vote threshold "lightness" of 226 (on a scale of 0 [black] to 255 [white]). That is, a vote oval that registered an average brightness of 226/255 or lower was counted as a vote. Approximately 50 ballots that were rejected by Ballot Browser were hand counted. These rejections were mostly due to excessive skew in the image.

The county and independent results agree to 0.01% on five of the six propositions. On proposition 1C, they agree to 0.02%.

As time permits, we may try to track down these (very minor) differences.

COUNTY - - - - - INDEPENDENT
#------- %-------- /// #------ %---------
1A
09661 37.44% / 09661 37.45%
16143 62.56% / 16138 62.55%

1B
10537 40.92% / 10537 40.91%
15215 59.08% / 15218 59.09%

1C
08751 34.00% / 08749 33.98%
16985 66.00% / 16995 66.02%

1D
08419 32.89% / 08422 32.89%
17179 67.11% / 17183 67.11%

1E
08151 31.83% / 08147 31.82%
17455 68.17% / 17457 68.18%

1F
20966 81.18% / 20965 81.17%
04861 18.82% / 04865 18.83%

May 19 ballot images now online

2009-06-13T11:39:00.000-07:00

Lower resolution ballot image "montages" are available now at http://www.humetp.org/may19montages . Zip files of the original scan collection will be available on higher bandwidth sites sometime soon, or you can purchase a DVD from Humboldt County's elections office.

There are 26,040 ballot images (a few of which represent accidentally scanned ballot backs) in 260 montage files.

Although the ballot count was approximately 22,000 in Humboldt's election night final report, the final count includes approximately 4,000 additional ballots, mostly vote-by-mail ballots that came in close to election day but were not counted by election day.

We've already performed a preliminary count of these images via Ballot Browser, and the preliminary Ballot Browser vote percentages match the current election office results to well within a tenth of a percent. We are now going over the approximately 150 images that Ballot Browser kicked out for manual inspection, and will present complete independent counts within a few days.

Less drama this time, huh?

May 19 Special Election

2009-06-04T10:04:00.001-07:00

Yes, the transparency project will be releasing ballot images from the May 19 Special Election real soon now. Volunteers have scanned 21,000 of the just under 22,000 ballots cast. Ballot images and at least one independent count will likely be available sometime next week.

MN audit report

2009-06-04T10:01:00.000-07:00

Citizens for Election Integrity Minnesota has released what looks to me like an extraordinarily well-done report: “Eyes on the Vote Count: Non-partisan observer reports of Minnesota's 2008 audit and recount" It can be downloaded at www.ceimn.org

Ballot Browser featured in April Python Magazine

2009-05-01T10:49:00.000-07:00

For the geek in us all,

Ballot Browser is featured on the cover of April's Python magazine.

http://pymag.phparch.com/c/issue/view/96

volunteers needed for May election

2009-04-21T07:43:00.000-07:00

The Election Transparency Project will be scanning the ballots cast in the May special election. Because many vote-by-mail ballots will be returned before election day, we will start scanning (but not counting) during the week before the election. Ideally, we will be finished a week after the election.

If you live in Humboldt, can volunteer, and have the technical skills to drive a car or flag a bus, please contact Mitch Trachtenberg via email at mjtrac sort-of-an-"a" gmail.com.

Sec of State public hearing

2009-03-25T15:04:00.001-07:00

There's been a bit of press around the California Secretary of State's public hearing into decertification of the GEMS version used in Humboldt County.

The big news is this: even current versions of the GEMS tabulation system don't note the deletion of decks of ballots.

Too many sources to link, but Kim Zetter's piece on Wired is a great start.

Ballot Browser update

2009-03-25T15:01:00.000-07:00

Version 0.52 of Ballot Browser is now available here.

CA SoS report on the Deck Zero Problem

2009-03-03T09:01:00.000-08:00

The California Secretary of State's office has released their report on the Deck Zero problem that was (re)discovered as a result of the Humboldt County Election Transparency Project. The link is here:

http://www.sos.ca.gov/elections/elections_vs_premier.htm

State considers decertifying election software

2009-02-22T07:58:00.000-08:00

The slow-motion crash continues....

http://www.times-standard.com/localnews/ci_11760632

State considers decertifying election software
Thadeus Greenson/The Times-Standard
Posted: 02/22/2009 01:27:29 AM PST

California Secretary of State Debra Bowen is considering withdrawing
the state's approval of the flawed Premier Elections Solutions
software that resulted in almost 200 ballots disappearing from
Humboldt County's final November election results.

Bowen's office announced that it will hold a public hearing March 17
at her Sacramento office to discuss the findings of its investigation
into the problem, and to take public comment on the possible
withdrawal of the state's approval of the voting system. Meanwhile,
the two other counties in the state that use the system are left
contemplating what its decertification would mean, and some election
advocates are lobbying for some punitive action to be taken against
Premier Elections Solutions.

"The public hearing is typically the last step in the process,"
Secretary of State spokeswoman Nicole Winger said. "It
offers one final opportunity for input, for the vendor to state their
side of the story and for the public to give comments or input as
well. It's a chance for everyone to be on the same page and see where
things stand."

The software error reportedly first came to Bowen's attention after
Humboldt County Registrar of Voters Carolyn Crnich informed her office
that the first-of-its-kind Humboldt County Election Transparency
Project had uncovered a discrepancy in the final vote tallies from the
county's November election.

The project found that 197 vote-by-mail ballots, which had
been scanned through vote counting machines, mysteriously disappeared
from the final ballot tally as tabulated by Premier Elections
Solutions' GEMS software.

The problem was traced to a programming error with the specific
version of the software used in Humboldt County -- GEMS version
1.18.19 -- a programming error that sometimes results in the first
deck of ballots scanned through the vote counting machine vanishing
without a trace from the final results.

Chris Riggall, a spokesman for Premier Elections Solutions, said in a
previous interview with the Times-Standard that the company had known
of the programming error since 2004. Saying the certification process
is too lengthy and time consuming to have had the software
re-certified, Riggall said Premier instead issued "work around" orders
by e-mail to its customers instructing them how to take steps to avoid
the problem.

The terse, 40-word e-mail, a copy of which was obtained by the
Times-Standard, tells the vendor's customers that it is "very
important that you follow these instructions," and urges them to
contact

the Premier's representatives with any questions.

In two other California counties using the same software -- San Luis
Obispo and Santa Barbara -- elections officials included the "work
around" orders into their written Election Day procedures. In Humboldt
County, then Elections Manager Lindsey McWilliams said he received the
e-mail, but failed to pass the information along to his boss at the
time, Crnich, or his successor, Kelly Sanders.

The transparency project that discovered the error passes every ballot
cast in an election through an optical scanner after it's been
officially counted. The ballot images are then placed online, along
with open-source software, created by volunteer Mitch Trachtenberg,
that allows viewers to sort the ballots by precinct and scrutinize the
vote as they see fit.

Trachtenberg and Crnich met last week with Bowen and members of her
staff to offer a demonstration of the project, and they said the
presentation went well.

Careful to say that Bowen has not yet endorsed the transparency
project, Winger said her office was certainly interested to hear about
the fledgling project.

"Secretary Bowen is always interested to hear the details of efforts
to make elections more transparent, auditable and accurate," Winger
said. "She certainly is open to ideas about how to accomplish that on
a secure, statewide level."

Reached Thursday, Riggall said Premier believes the version of its
GEMS software in question should be taken out of circulation in
California.

"We completely agree that that version of GEMS should not continue to
be used," he said. "We absolutely have no problem and certainly would
concur were the secretary to take that action directing counties to
use a later version."

In Humboldt County, the Board of Supervisors has already approved a
plan for the county elections office to switch to Hart InterCivic
equipment. Crnich is currently working with Bowen's office to come up
with a plan to fund the switch, which will then come back before the
board for final approval.

But, the possible decertification of the version of GEMS currently
used in Santa Barbara and San Luis Obispo is raising some questions
there for elections officials, especially with the state calling a
special election in May. Officials in both counties said they are
still looking into how they would proceed if the approval of their
voting systems were to be withdrawn.

Meanwhile, some in the election watchdog community are pushing for
either the Secretary of State's Office or the federal Elections
Assistance Commission to pursue punitive actions against Premier,
which they say knowingly kept elections systems in place that had
unacceptable error rates.

Riggall said Premier has done nothing disingenuous, and notified its
customers immediately upon discovering the error in its vote counting
system, carefully instructing them how to "work around" the problem.

"I don't see that there is anything -- absolutely nothing -- in how we
have handled this issue going back several years that has been
disingenuous," Riggall said.

Trachtenberg, on the other hand, said he thinks it would be
appropriate for the Department of Justice or the Attorney General's
Office to investigate the matter.

"I think that would be totally appropriate," Trachtenberg said. "They
sold software that doesn't work."

Thadeus Greenson can be reached at 441-0509 or tgreenson@times-standard.com.

independent count now available in CSV form

2009-01-10T13:49:00.000-08:00

I've just uploaded contest-by-contest CSV files (Comma Separated Variables -- it's a spreadsheet thing) for many of Humboldt County's November 2008 contests. They are at http://www.mitchtrachtenberg.com/Nov2008. The files are based on the output from Ballot Browser operating on scans done by the volunteers of the Humboldt County Election Transparency Project and are as unofficial as anything can be. The file format is explained at the link.

update on Humboldt's Nov 2008 results

2009-01-08T16:57:00.000-08:00

Joint Statement on the November 2008 Humboldt County Election Results

Carolyn Crnich,
Humboldt County Clerk and Registrar of Voters

Mitch Trachtenberg,
Humboldt County Election Transparency Project volunteer

January 8, 2009

As we've compared the results from Humboldt County's official count with the independent count Mitch has conducted with his Ballot Browser independent vote counting software, we've found two additional issues.

First, the Election Transparency Project had scanned the front side of 63 ballots twice (once upside down); these duplicate scans will be removed from Ballot Browser's counts.

Second, the Elections office appears to have scanned 57 ballots into the Diebold GEMS system twice -- these duplicates need to be removed from the GEMS results.

The numbers from the two systems are now extremely close, though not identical.

We believe many of the remaining variations may be a result of differing vote sensitivity between the Diebold system and Ballot Browser, with Ballot Browser's totals approximately 0.05% higher than those from the Diebold system (approximately one added vote per 2,000 counted vote opportunities).

The variations that remain do not affect the outcome of any races.

November 2008 Ballot Images are Online

2008-12-15T10:06:00.000-08:00

The Transparency Project's scanned images of Humboldt County's November 2008 ballots are now available online. Thank you to Streamguys, especially Jonathan Speaker and Andy Jones, for helping us get the bits out of our beautiful low-bandwidth region. And thank you to EARC for offering the hosting space.

The UC Berkeley Election Administration Research Center is hosting the images in collection form, as a series of approximately one gig files and accompanying GPG signatures. The site is momentarily password protected; if it is still password protected when you read this, we've been given permission to give the case-sensitive user/password combination of "download/Down2disc".

The ballots will also be hosted as individual images (and perhaps as collections as well) at Parke Bostrom's ETP site.

We expect the ballots to become available at other sites as well.

The latest version of Ballot Browser software is also available. This location will point to updates as they appear.

Time-Standard followup, and more

2008-12-07T17:54:00.001-08:00

Thadeus Greenson has written a follow-up to his story from Friday. It appeared in our local newspaper, The Eureka Times-Standard, on Sunday.

http://www.times-standard.com/localnews/ci_11161383

I've written my own commentary, it's at my web site:

http://www.mitchtrachtenberg.com/ourvotes.html

One of the founding volunteers, Tom Pinto, who works at the Humboldt County District Attorney's office, has put up a web site for the project at http://humtp.com. This site has a lot of history and pictures, in addition to an overview of the project.

The story is also getting a fair amount of exposure in the "election integrity community," whatever that is. It's been picked up by Brad Friedman's http://www.bradblog.com.

The Peter B. Collins national radio show had Registrar Carolyn Crnich on its Friday program.

Where's that counting software you're talking about?

2008-12-07T16:57:00.000-08:00

Here's the latest development release of the vote counting software I've been developing:
http://www.humetp.org/newbbhum2008.zip

Here's the matching "signature"
http://www.humetp.org/newbbhum2008.zip.asc

The software hasn't yet been packaged up for full public release but it is, after all, open source, which means I can't hold onto it forever.

Software glitch yields inaccurate election results

2008-12-05T07:39:00.000-08:00

Here's a link to coverage of the discrepancy the project found, possibly exposing a four year old flaw in Premier Election Systems software. The story is by Thadeus Greenson of the Eureka Times-Standard.

http://www.times-standard.com/localnews/ci_11145349

We Scanned More Ballots than were in the Official Results

2008-12-04T18:31:00.000-08:00

Despite accurate work by our elections department, their official count was missing 197 ballots from a single precinct. Humboldt's version of GEMS appears to drop the zero'th ballot deck under certain circumstances.

Statement by Mitch Trachtenberg,
developer of Ballot Browser, open source vote counting software

Email: mjtrac@gmail.com

Our votes are too important to be counted by secret code running on proprietary machines.

Since 2000 and before, many people have warned that the vote counting machines might not be doing exactly what they say they do.

Today's results demonstrate that the warnings are correct. A stack of 197 valid ballots was not included in Humboldt County's official count, despite the accurate work of Humboldt County's elections department.

As far as I understand, this appears to be due to an error in Premier Election System's software... an error Premier -- formerly Diebold -- may have known about.

Our Election Transparency Project completed its scan of all ballots late on Sunday, and it was clear that our count of ballots scanned did not match the count of ballots in the official results. By counting votes with Ballot Browser, open source software running on the open source platform Linux, we were quickly able to localize the problem to precinct 1E-45. This precinct turned out to include the first deck of ballots run through the Premier system, and it was not included in the results produced by Premier's software.

Has this happened in other counties? How can we know?

Our votes are too important to be counted by secret code.

Scanning is complete

2008-12-03T06:51:00.001-08:00

The scanning process has been completed. We expect to be releasing the images within a few days. We are now doing preliminary counts on the scanned ballots, and plan to release them once we've had a chance to look over the counts.

46,000 ballots and counting

2008-11-18T12:32:00.000-08:00

As of this morning, we've scanned approximately 46,000 ballots. The elections department is still counting absentee ballots delivered to precincts, provisional ballots, and others. We'll hold off on the last of the scanning until the elections department has completed their process with these ballots.

We've found that a team of two, using our Fujitsu production scanner, can process about 1,000 ballots per hour. Because we can only imprint a serial number on one side as we scan a ballot, we've been running each ballot through the scanner twice. That is, we could probably operate at closer to 2,000 ballots scanned per hour if we were willing to use the scanner's duplex output and live with one imprint per ballot.

November election update

2008-11-15T10:40:00.000-08:00

As of the end of Friday, November 14th, we've scanned 62,000 sides, or 31,000 ballots. That's roughly half of this election's ballots. We are scanning on Saturday and hope to complete scanning by midweek, at a rate of approximately 1,000 ballots per hour. Preliminary vote counts from the scanned images should be available by the end of the week.

November election

2008-11-09T19:21:00.000-08:00

We started scanning the election of November 4th on Friday, November 7. As of the end of Saturday November 8th, we had scanned approximately five thousand ballots, about eight percent of the total. We'll continue scanning throughout the coming week. It should take approximately 65 hours of scanning to complete the election. We are only given ballots to scan after the registrar has reconciled a precinct's ballot count with its signature log. We hope to have ballot images available before Thanksgiving.

Can't digital images be faked?

2008-07-18T11:29:00.000-07:00

Yes, it's easy to fake digital images.

Here's the process by which we ensure that isn't happening.

Our scanner is not a special piece of elections equipment, it is a general purpose office scanner.

The scanner will be controlled by a computer that is running an open source operating system, (Debian Linux "Etch"). All computer programs that control the scanner are available for inspection by anyone who would like to see how they operate. The counting and sorting software will always be available for inspection as well.

The Linux system is not configured for networking.

The ballot images from the scanner are archived into a single large file. This large file is then digitally signed using a technology known as public key encryption. This generates a small signature file, which can be printed or emailed. Public key encryption works with pairs of keys -- one key is secret and the other is public. The secret key is used to sign the ballot file. To sign the ballot file, a person needs to be in the room with the scanner computer, have the password to the elections account, and know the "passphrase" that unlocks the signature program. The public key is available on the Internet.

Anyone wanting to ensure that the ballot file they've received has not been altered can run a free program called GPG ("GNU Privacy Guard") to validate that the ballot file and signature go together, and that the signature was generated by someone with access to the elections office private key. A change of even one bit in the entire file will prevent the file from being validated.

This gives us a very high degree of confidence that, if you validate our file with GPG, the images
we scanned are the images you get.

Each ballot is imprinted with a unique number just as it enters the scanner. This number is part of the scanned image, and can be used to locate the paper ballot associated with the image. If any questions about our images were to arise, a random sampling of the images, checked against the ballots themselves, would allow anyone to confirm that the images we provide are actual images of the ballots themselves.

Ballot Browsing program

2008-07-08T10:19:00.001-07:00

Here's a screen shot of a ballot browsing program I'm developing for the project. A download should be available soon.

The program is built in Python and uses Tkinter, Pysqlite and the Python Imaging Library (PIL); everything is free and open source, and can run on Windows or Linux.

In addition to showing the ballots, this program can count votes for a precinct, and show how it decided which voting ovals were marked, showing the user the oval in question.

In this screenshot, the program has overlaid the selected ballot with rectangles showing which areas the program is counting, and which candidates get votes when the ovals in those rectangles are filled in. Under the menubar, the program has listed the winners on this particular ballot.

The program can use ballots installed on an individual's computer from the project's DVDs, or can go out and retrieve individual ballots over the internet.

June election images are now online

2008-07-07T16:33:00.000-07:00

An unofficial set of ballot images from the Humboldt County June 2008 election is now available at http://hum.dreamhosters.com/etp

The ballot images were uploaded and are being hosted by project volunteer Parke Bostrom.

These images are available sorted by individual precinct.

Humboldt County Election Transparency Project

2008-07-03T05:51:00.000-07:00

In recent years, citizens throughout the United States have expressed concern about the legitimacy of machine counts. Voting systems that do not provide a paper audit trail require that citizens trust the company that has provided the voting system. Voting systems such as that used in Humboldt County, which use optically scanned paper ballots, do leave an audit trail of all cast ballots. This audit trail becomes far more valuable if it is actually used to verify the count.

The Humboldt County Election Transparency Project aims to provide images of each counted ballot, so that any person or organization wishing to do an independent count will have access to a complete set of ballot images.

Project members are writing our own software to count and sort the images, and this software is open-source, meaning that our programs' "source code" will be available to anyone. Anyone who wants to rebuild the programs and run them on their own computer can do just that without buying a thing. Also, by November, we expect that we will be scanning using Linux, an open-source computer operating system. And we'll be using open-source code to run the scanner at the county Elections office. Finally, the scanner itself is not special election-related hardware -- it's just a high-speed, general purpose office scanner.

The project was initiated by Humboldt County County Clerk and Registrar of Voters Carolyn Crnich, former Green party Presidential candidate David Cobb, local citizens Kevin Collins and Parke Bostrom, and Tom Pinto of the Humboldt County District Attorney's office. The custom scanning and counting software, along with the technical procedure used by the project, has been developed pro bono by Humboldt-based freelance programmer Mitch Trachtenberg, who may be contacted at mjtrac@gmail.com. The scanning software will use the SANE ("Scanner Access Now Easy") protocol, and is built on top of the SANE command line program "scanimage". These programs run on top of the Debian Linux Etch release, downloaded from a debian mirror in June 2008. All software used is available for download free of charge. (Download site TBA.)

Our first run was on the June 2008 election, and we have scanned the roughly 30,000 ballots cast. This scan was run using Microsoft Windows to drive the scanner, which captured the images at a resolution of 150 dots per inch. We used Windows while we wait for modifications to the SANE scanner driver which will allow it to control the imprinter that prints serial numbers on the scanned ballots. The images have now been provided to Dr. David Dill of Stanford, Joseph Lorenzo Hall of Berkeley, and Bev Harris of Black Box Voting. We expect they will be available soon for anyone to view or download. The images are also available on 3 DVDs from the Humboldt County elections office.